Kozminski Techblog

Reading Time: 3 minutes

TikTok, the popular short-video sharing platform has become a global phenomenon in the past couple of years. It already has over a billion active users. It has definitely become a very powerful and helpful tool for creators and influencers to share their content or connect with new spectators. An extensive and advanced recommendation algorithm played a huge role in TikTok’s success. 

TikTok’s recomendation algorithm uses a variety of factors to choose what videos to show to every single user. The algorithm takes into account a vast range of data such as the type of content that the particular user watches, general users behaviour and their interactions with other users. 

One of the most important factors that the algorithm considers is the user’s viewing history. If a user watches a video all the way through, leaves a comment or a like, the algorithm will remember that and show the user more videos like that, as they might be interested in them. This ensures that a user is being show only the type of content that they like and enjoy. The algorithm also show the most popular videos among other users providing a sense of community and social interaction. 

User’s interaction with other users is also a very significant factor. When a user gets involved with a particular user’s content by liking, commenting or sharing, the algorithm notes this behaviour and then starts showing more clips from that particular user. This consideration of a user’s interactions with others, like the previous factor, also promotes a sense of community allowing users to be shown content of a user they have previously interacted with.

Sometimes while scrolling through TikTok you might see a video that has been posted a while ago. That’s because the clip has been nothing but popular and trending since it was released, meaning TikTok is suggesting content for you to discover, which you have previously missed, but you might be interested in. This is based on the assumption that you might like something that has gone viral and was liked by many other users. 

This algorithm’s key strength is the ability to personalise content for each individual user. To do it I gathers a vast amount of data and machine learning techniques to learn and understand the preferences of each user. As the algorithm learns about the types of videos each user is likely to like, it adjusts the content according to those preferences. This keeps users engaged with the app and keeps them coming back every now and then. 

However, while the algorithm has come in for praise for the ability to personalise and improve users experience it has also faced criticism for its potential to be manipulated to promote certain types of content over others or to certain types of content to some users based on their demographic information. 

What’s more, the algorithm prioritises content that gets the most attention, likes, views. This sometimes could lead to the spread of misinformation or harmful content. Videos with the biggest amount of attention are much more likely to be shown to more people, even when the information on the video is false of dangerous in some way. 

In summary, TikTok’s recommendation algorithm is a very valuable feature, that aligns the content that users watch with their interests. However its important for the company to make sure that the algorithm is not misused or overused in any way, for instance to spread false information.  

sources:

–https://newsroom.tiktok.com/en-us/how-tiktok-recommends-videos-for-you

-https://sproutsocial.com/insights/tiktok-algorithm/

Reading Time: 3 minutes

Electric vehicles (EVs) have been growing in popularity in recent years because of the potential they have in reducing carbon emissions and improving air quality. Unlike typical gasoline-powered vehicles, which release toxic gases into the atmosphere, electric ones release no fumes and can help to fight climate change. Although the price of an EV may be higher, the money you pay will surely pay off after time and they have several advantages which make them a better and more environmentally friendly choice. 

It is said that electric automobiles produce no emission at all and that would have been true statement, because they have no tailpipes, if not the fact, that we cannot exclude the process of making and charging the machines, which unfortunately also creates some emissions. Even so, vehicles that are electric will never produce as much carbon emissions as the gasoline-burning cars, which is a big advantage of the EVs over classic gas-powered cars, which are mostly responsible polluting the air in many cities in the world. 

The creation of a massive lithium-ion batteries is probably the main source of EV emissions. Cobalt or lithium, minerals which are the essentials to create such battery must be mined and then heated. Those processes require the use of fossil fuels. Tesla Model 3 now uses a 80 kWh lithium-ion battery. Building a battery that big, to this exact car model, forces us to create about 12 metric tons of CO2.  From this it follows, that by producing a new Electric Vehicle, we also produce about 80% more emissions than if we produced a classic gas-powered car. However as we know the technology constantly improves, so the emissions associated with manufacturing batteries and general EV production are very likely to drop. 

The situation gets much better when Electric vehicles leave their factory. The biggest source of EV emissions then becomes the energy that is used to charge the batteries located inside the cars. The amount of created pollution varies on for example the energy that is used to charge them. Studies found that if the power source are fossil fuels, coal or gas, the average emission created by a fully electric car will only be about 200 grams of CO2 for every driven mile, while gasoline-powered cars create 350 grams of CO2. However if the electricity comes from solar, wind, or other renewable sources then the difference in emissions is beginning to become significant. Either way, in most cases choosing en electric vehicle can still be a better and wiser option than gasoline-powered automobile. 

By 2050 classic gasoline vehicles are supposed to be even more efficient, whereas the electric vehicles are expected to become their best possible version. From more than 350 grams of CO2 per mile, fuel-powered cars pollution is supposed to decrease to about 225 grams of CO2 by that time. EVs battery, however, is able to drop 125 grams of CO2 and maybe even to only 50 gram of CO2 per mile in that same timeframe. 

I would not say it is correct to say that electric vehicles are definitely better for the climate than fuel cars. Even though EVs have a huge potential to reduce the air pollution, this all not only depends the car, but on more than one factor including the source of electricity that is used to charge it and the process of its making. 

Sources:

-https://www.transportenvironment.org/discover/how-clean-are-electric-cars/

-https://climate.mit.edu/ask-mit/are-electric-vehicles-definitely-better-climate-gas-powered-cars

-https://www.energysage.com/electric-vehicles/101/pros-and-cons-electric-cars/

-https://www.edfenergy.com/for-home/energywise/electric-cars-and-environment

Reading Time: 3 minutes

I’m sure that many terrible things can happen within just one year, but for Apple Music users and not only, December is when it gets „the worst”. It’s the time of year, when we all have to see every single person’s Spotify wrapped. It literally takes over the Internet. It takes over Instagram, Twitter, TikTok and sometimes even Facebook. 

For users of Apple Music (myself included), it’s like being excluded from   the society for some time. We cannot evade, and still we cannot participate. We are compelled to observe. Unhappily for us, Spotify has made Wrapped a cultural masterpiece. 

In 2022 its has changed, because Apple has prepared a Spotify Wrapped competitor called Apple Music Replay, which has technically existed since around 2018, but it only used to create a playlist of the 100 most listened to songs of a given year. The stats that you now get from Replay still aren’t as great and „social mediable” as the ones you get from Spotify. Nonetheless you get the basics like top songs, artists or albums. You can obviously live with this little information about the music you’ve been listening to for the past year, but the problem appears when you are trying to share it. Its surely not easy. That is the main reason why Spotify Wrapped is still much better than Replay. What’s more, you still don’t get any summary from Apple. I’m sure most of you have seen Spotify users sharing their top songs, top genres and minutes spent on listening. Apple Music simply doesn’t have that on just one card. You can still do that by tapping through the whole Replay and taking screenshots of each section. Screenshots, because you don’t even have a button with which you can share it. Then another problem emerges, which is that Apple apparently doesn’t want you to share your Replay to anyone, because it is presented in the least „screenshotable” form. Namely, top songs and top artists are shown in long lists of 10 and the most you can fit on one screenshot is 7, which makes it quite evident that you’re not showing the full list. 

I absolutely get that those complaints are just small problems of the big world, but Apple Music clearly didn’t make Replay for people to share it with each other and  I think that social media sharing is what made Spotify wrapped such a cultural masterpiece. You can fight me about that one, but in my opinion music is definitely a fundamental part of our lives and Spotify wrapped isn’t just about that. It’s a brilliant solution to share parts of our lives, of us. It could be also a new way to get to know people we like. Replay just needs to be better. Apple Music would save so many more users if it could just create a product that actually measures up to Spotify’s wrapped. I’ll be honest, it’s tempting to transfer from Apple Music to Spotify just to share something so coherent, consistent and aesthetically pleasing that is available only once a year. 

Sources:

-https://www.apple.com/newsroom/2022/11/apple-music-launches-new-replay-experience-reveals-2022s-top-charts/

-https://newsroom.spotify.com/2022-11-30/everything-you-need-to-know-about-2022-wrapped/

And also my own experience as a long-time Apple Music user

Reading Time: 2 minutes

Artificial intelligence is revolutionising in many sectors. One of them is sport which is being taken to the whole new level thanks to its multiple systems focused for example on improving the performance of athletes and coaches. AI in sports is definitely a new reality that is having a significant impact on many of its aspects.

Those systems have been created to change the way things are being done in sport. In a sector that is increasingly moving towards professionalisation in areas. Those types of systems provide a tremendous capacity in terms of efficiency and performance. There are many different areas in the sports industry where AI is beginning to play a fundamental role. Some of these include: Marketing, medical services, performance of athletes, scouts, sports refereeing or even automated journalism.

Hawk-Eye could be a great example of such system. It helps resolving controversial moments during the game. Cameras record, temporal and visual data, which the system automatically processes. Giving such system detailed information about the rules of the particular discipline or the playing area will create a tool that will be able to spot any abnormalities in real time. For instance, a ball going out of bounds. This system was used to support referees’ job, but on Grand Slam Australian Open tournament in 2021 for COVID reasons, a vast amount of Hawk-eye cameras completely replaced the line referees, instantly calling out balls, as well as decisions on serves hitting the net. This system is also able to archive data, which can be later used to create statistics or analyze players’ behaviour.  

Another helpful system can be SAS, which is used, by one of the analytics companies, SciSports. The company supports clubs in finding players for specific positions and developing tactical and training objectives. SciSports works with elite football clubs in the whole world but mainly in Europe to help them select the best players and build optimal teams. SAS system analyses more than 1,500 matches in 210 leagues per week, using machine learning to provide the estimated value, condition and potential of more than 200,000 players. The company uses SAS Visa platform and their own BallJames camera system to create a 3D video of the recently played matches. SAS enables a very detailed analysis of players’ performance. All the collected data can be used to identify underrated players or new rising stars. All that by comparing statistics of a particular player with every other player in their league.

To conclude, according to the latest AI research, a great advantage of using such systems in sport is that they apply a certain objectivity to the analysis of the game without the need to rely on a human expert who might get different results depending on factors such as experience, context or knowledge.

Also one of the biggest advantages of artificial intelligence systems in sport is the large amount of data that can be processed, since, logically, machines have a greater capacity to store and evaluate huge amounts of information, which makes them a better analysts than human beings could ever become. 

Sources: 

–https://www.sas.com/en_us/23289/2323/sas-viya-on-azure-marketplace.html

–https://bodyler.com/innovations/how-does-the-hawk-eye-system-work/